Fix encoding errors on Windows with certain languages.

[realRurichan]

Description

When running hyfetch in Windows systems installed from images with certain languages (such as Simplified Chinese and Traditional Chinese), an encoding error may occur.

Error: 'utf-8' codec can't decode byte 0xb1 in position 12: invalid start byte
Traceback (most recent call last):
  File "C:\Users\ruri\AppData\Local\Programs\Python\Python311\Lib\site-packages\hyfetch\main.py", line 389, in run
    asc = get_distro_ascii() if not args.ascii_file else Path(args.ascii_file).read_text("utf-8")
          ^^^^^^^^^^^^^^^^^^
  File "C:\Users\ruri\AppData\Local\Programs\Python\Python311\Lib\site-packages\hyfetch\neofetch_util.py", line 315, in get_distro_ascii
    det = distro_detector.detect(distro or get_distro_name())
                                           ^^^^^^^^^^^^^^^^^
  File "C:\Users\ruri\AppData\Local\Programs\Python\Python311\Lib\site-packages\hyfetch\neofetch_util.py", line 336, in get_distro_name
    return run_neofetch_cmd('ascii_distro_name', True)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Users\ruri\AppData\Local\Programs\Python\Python311\Lib\site-packages\hyfetch\neofetch_util.py", line 296, in run_neofetch_cmd
    return check_output(full_cmd).decode().strip()
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
UnicodeDecodeError: 'utf-8' codec can't decode byte 0xb1 in position 12: invalid start byte

This is because the system name returned by the "wmic.exe os get Caption" command uses the "ISO-8859-1" encoding, which cannot encode Chinese characters. Regardless of how it is decoded, it will be garbled. Hyfetch itself uses utf-8 for decoding, but the error occurs due to inconsistent encoding.

Features

I switched to using the registry to obtain the system name in order to avoid compatibility issues on Windows systems installed from images with certain languages.

Issues

TODO

[hykilpikonna]

Great work, thanks!

[slackingfred]

I want to add something about the cause of the original issue.

The wmic command uses a tricky check to determine which encoding to use: UTF-16 if writing to a file, or the OEM codepage (system locale) if writing to a terminal. (see here) When using bash's $() syntax to capture command output, wmic behaves as if stdout is a terminal. I can confirm that the output is in CP936, aka GBK, on a Simplified Chinese system.

I tried to integrate the chcp command to get UTF-16 output (see here), but it didn't work.

I think we need to replace every wmic command in neofetch to avoid further issues. Or implement a function to get the system locale, and interpret the output accordingly.